Re: bug in bcachefs -> bio_copy_data_iter

[Christoph Hellwig <hch@lst.de>]

On Wed, Nov 10, 2021 at 08:02:06PM +0800, Coly Li wrote:
>> looks like it should be functionally equivalent), but clearly I'm missing
>> something... wonder if there might be some relation to the bug you guys hit in
>> bcache with bvec_virt.
>>
>> Any ideas?
>
> I experience similar one in bcache code during my recent development,

Can you try this patch?


diff --git a/block/bio.c b/block/bio.c
index 4259125e16ab2..ac29c87c6735c 100644
--- a/block/bio.c
+++ b/block/bio.c
@@ -1336,10 +1336,12 @@ void bio_copy_data_iter(struct bio *dst, struct bvec_iter *dst_iter,
 		struct bio_vec src_bv = bio_iter_iovec(src, *src_iter);
 		struct bio_vec dst_bv = bio_iter_iovec(dst, *dst_iter);
 		unsigned int bytes = min(src_bv.bv_len, dst_bv.bv_len);
-		void *src_buf;
+		void *src_buf = bvec_kmap_local(&src_bv);
+		void *dst_buf = bvec_kmap_local(&dst_bv);
 
-		src_buf = bvec_kmap_local(&src_bv);
-		memcpy_to_bvec(&dst_bv, src_buf);
+		memcpy(dst_buf, src_buf, bytes);
+
+		kunmap_local(dst_buf);
 		kunmap_local(src_buf);
 
 		bio_advance_iter_single(src, src_iter, bytes);